【摘要】 目的:探讨基于磁共振表观扩散系数ADC图纹理分析技术在鉴别肝癌与血管瘤中的可行性。方法:回顾我院56例行1.5TMR扫描患者(肝癌31例,35个病灶;血管瘤25例,27个病灶)。将ADC图像导入Omni-kinetics(OK)软件得到未经滤过的原始细纹理图像,并测量感兴趣区内的纹理参数(能量值、熵值、偏度值和峰度值)。测量结果进行组内相关系数检验(ICC)一致性进行检验,比较两组病灶各参数值并进行两独立样本t检验及受试者操作特征(ROC)曲线分析。结果:两名观察者所测得各参数值一致性良好(ICC>0.75),肝癌及血管瘤的能量值、熵值、偏度值和峰度值分别为:0.03±0.05、6.82±0.92、0.12±1.35、4.43±4.55;0.12±0.15、6.02±2.21、0.69±2.01、6.98±12.81,血管瘤的能量值高于肝癌(P<0.05),而熵值、偏度值和峰度值两组之间没有明显的统计学差异(P>0.05)。ROC曲线分析显示能量值的阈值为0.04时,曲线下面积、敏感度和特异度分别为0.827、71.0%和79.2%。结论:磁共振ADC图的纹理分析技术对鉴别肝癌和血管瘤有一定价值。更多还原

【Abstract】 Purpose: To explore the value of texture analysis based on apparent diffusion coefficient(ADC) mapping of magnetic resonance diffusion weighted imaging(DWI) for differentiating hepatocellular carcinoma(HCC) and hepatic hemangioma(HHA). Methods: Fifty-sixth patients, who had undergone MRI exam on a 1.5 T MRI scanner, were enrolled in this study. All patients were classified as HCCs(31 patients with 35 lesions) or HHAs(25 patients with 27 lesions) group. The ADC images were imported into Omni-kinetics(OK) software and the parameters of ADC texture were calculated automatically with texture and histogram algorithm. The parameters of ADC texture including energy,entropy,skewness and kurtosis were derived from ADC images.The values of energy,entropy,skewness and kurtosis were measured and recorded.Intraclass correlation coefficient(ICC)test was analyzed to test the consistency of two observers.The diagnostic value of ADC texture parameters was examined with independent samples t-test.The ROC curves were used to analyze the texture parameters with statistical significant and the diagnostic threshold was determined.Results:The ICC values of the texture parameters were all greater than 0.75,exhibiting an amenable consistency.For HCCs and HHAs,the values of energy,entropy,skewness and kurtosis were 0.03±0.05,6.82±0.92,0.12±1.35 and 4.43±4.55 vs.0.12±0.15,6.02±2.21,0.69±2.01 and 6.98±12.81,respectively.The energy values of HHAs were significantly higher than that of the HCCs(P<0.05).While other values were not shown with significant difference(P>0.05).When the threshold of energy value was set as 0.04,the AUC,sensitivity and specificity were 0.827,71.0%and 79.2%,respectively.Conclusion:Texture analysis based on ADC mapping of MR-DWI is helpful in differentiating the HCCs from HHAs.更多还原